Re-seed random ID generator after os.fork()

[alexmojaki]

os.fork() copied the SeededRandomIdGenerator state so that the parent and child processes would generate the same pseudorandom trace/span IDs, which is very bad. Demo script that this PR fixes:

import os
import logfire

logfire.configure()

logfire.info('log', pids=(os.fork(), os.getpid()))

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 100.00%. Comparing base (e793014) to head (73806bc).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff            @@
##              main      #644   +/-   ##
=========================================
  Coverage   100.00%   100.00%           
=========================================
  Files          135       135           
  Lines        10681     10683    +2     
  Branches      1465      1466    +1     
=========================================
+ Hits         10681     10683    +2     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[phil65]

Just fyi, this broke logfire for me (using windows):

File "E:\llmling.venv\Lib\site-packages\logfire_internal\utils.py", line 378, in post_init
os.register_at_fork(after_in_child=self.random.seed)
^^^^^^^^^^^^^^^^^^^
AttributeError: module 'os' has no attribute 'register_at_fork'

[alexmojaki]

Thanks @phil65 for reporting, released another fix: https://github.com/pydantic/logfire/releases/tag/v2.6.2